Water deficits accelerate ripening and induce changes in gene expression regulating flavonoid biosynthesis in grape berries
- First Online:
- Cite this article as:
- Castellarin, S.D., Matthews, M.A., Di Gaspero, G. et al. Planta (2007) 227: 101. doi:10.1007/s00425-007-0598-8
- 1.2k Views
Water deficits consistently promote higher concentrations of anthocyanins in red winegrapes and their wines. However, controversy remains as to whether there is any direct effect on berry metabolism other than inhibition of growth. Early (ED) and late (LD) season water deficits, applied before or after the onset of ripening (veraison), were imposed on field grown Vitis vinifera “Cabernet Sauvignon”, and the responses of gene expression in the flavonoid pathway and their corresponding metabolites were determined. ED accelerated sugar accumulation and the onset of anthocyanin synthesis. Both ED and LD increased anthocyanin accumulation after veraison. Expression profiling revealed that the increased anthocyanin accumulation resulted from earlier and greater expression of the genes controlling flux through the anthocyanin biosynthetic pathway, including F3H, DFR, UFGT and GST. Increases in total anthocyanins resulted predominantly from an increase of 3′4′5′-hydroxylated forms through the differential regulation of F3′H and F3′5′H. There were limited effects on proanthocyanidin, other flavonols, and on expression of genes committed to their synthesis. These results demonstrate that manipulation of abiotic stress through applied water deficits not only modulates compositional changes during berry ripening, but also alters the timing of particular aspects of the ripening process.
KeywordsAnthocyaninBerryFlavonoidVitis viniferaWater deficitsWine
Days after anthesis
Flavonoids are regarded as one of the most important determinants of quality in red grapes and wines. Color and taste of red wines are strongly related to the amount of anthocyanins, flavonols, and proanthocyanidins. Moreover, in recent years some flavonoid compounds (anthocyanins, proanthocyanidins, and flavonols) have attracted additional attention for their potential health benefits (e.g., Cos et al. 2004; Thole et al. 2006). Vine water status affects the accumulation of flavonoids (Matthews and Anderson 1988; Kennedy et al. 2002), and regulating vine water deficits is considered a tool for increasing flavonoid content and improving quality in red winegrapes (Matthews et al. 1990; Kennedy et al. 2002).
It remains unclear whether water deficits alter the pertinent biosynthetic pathways or high concentrations arise from the high sensitivity of berry growth to water deficits. Matthews and Anderson (1989) and Matthews et al. (1990) showed that the growth of berries was inhibited more and the concentration of flavonoids in fruit and resultant wine increased more when water deficits were imposed before veraison than after veraison. Based on the observation of similar flavonoid content per berry between treatments, Kennedy et al. (2002) concluded that postveraison water deficits only inhibited fruit growth. Roby et al. (2004) dissected the effects of berry size and of low vine water status on flavonoid concentration to show that there are effects of vine water status on fruit composition that are independent of the inhibition of berry growth. However, this analysis also indicated that the fruits developed during water deficits had relatively more skin, the predominant tissue of flavonoid biosynthesis, than control fruit. Profiles of gene expression combined with metabolite data should clarify whether or not water deficits differentially promote the biosynthesis of flavonoid compounds. The Kennedy et al. (2002) and the Roby et al. (2004) studies involved a single water deficit treatment that developed after veraison. Therefore, in this study we report the effect of the timing of water deficits on: (1) berry maturation, (2) the accumulation of anthocyanins, proanthocyanidins and flavonols, and (3) the expression of flavonoid pathway genes in berry skin.
Materials and methods
Field plots and physiological measurements
Field experiments were conducted during 2006 in a commercial vineyard of R.H. Phillips Winery in the Dunnigan Hills American Viticultural Area (Esparto, CA, USA) (38°45′N, 125°57′W). Grapevines of “Cabernet Sauvignon” clone 337 grafted onto rootstock 140R were planted at 2.43 m × 1.83 m spacing (2,242 vines per hectare) in east–west oriented rows and trained to bilateral cordons.
Three irrigation treatments were established. Irrigation was applied to control (C) vines in order to maintain midday leaf water potential (ψl) between −0.9 and −1.2 MPa throughout the season. Irrigation was cut off to ED vines from fruit set till two days before the end of veraison (77 days after anthesis, DAA); then vines were irrigated at the C rate till harvest (133 DAA). LD vines were irrigated at the same rate as C vines till 77 DAA when irrigation was cut off. Plant water status was monitored weekly by measuring midday leaf water potential (ψl) of fully mature, sun-exposed leaves, as described previously (Matthews et al. 1987a). Water was supplied by an underground drip irrigation system at approximately 50 l per vine per week. No rain occurred after fruit set.
Each treatment was replicated four times in 0.5 ha plots dispersed over a 36 ha vineyard according to a randomized complete block design. Each plot consisted of five rows separated by one buffer row. Samples were collected from ten vines located in the central row of the plots.
Berries were sampled ten times between 27th June (33 DAA) and 5th October (133 DAA), when the grapes were harvested. In order to minimize the effect of cluster exposure to direct sun, and to use the fruits exposed to a uniform diffuse light intensity, berries were collected from the north side of the row and only from the sides of clusters not directly exposed to the sun. Two sets of 20 berries were collected from each plot. The first set was used for measurements of berry weight and soluble solids, whilst the second was used for flavonoid analyses and studies of gene expression. During veraison, additional samples of green berries were collected separately from the same clusters and analyzed separately. Data were then expressed as weighted average.
Samples for soluble solids measurements were weighed and pressed. Total soluble solids were quantified in juice with a hand held refractometer and expressed as °Brix. The corresponding sample of 20 berries was peeled and skins were immediately frozen in liquid nitrogen. After being finely powdered, one aliquot of 0.1 g was used for anthocyanin and flavonol analyses, one aliquot of 0.1 g was used for proanthocyanidin analysis, and one aliquot of 0.3 g was used for RNA extraction.
Anthocyanins and flavonols were extracted in 1/10 (w/v) skin/solvent suspension of 50% methanol in water for 20 min in an Ultrasonic Cleaner bath (Branson 2000). After centrifugation at 13,000 × g for 15 min, samples were injected into an Agilent 1100 Series HPLC system equipped with a SGE Wakosil II C18 RS column (3 μm, 150 × 4.6 mm) and a guard SGE Wakosil II C18 RS cartridge. Gradient conditions were the same as reported by Bogs et al. (2005). Total anthocyanins were expressed as mg L−1 of malvidin 3-glucoside and included monoglucoside, acetyl-glucoside, and p-coumaroyl-glucoside fractions. The composition of monoglucoside anthocyanins was used for the calculation of the proportion of 3′4′5′-OH/3′4′-OH derivatives.
Two major flavonol peaks, corresponding to quercetin-3O-glucuronide and quercetin-3O-glucoside peaks, were identified by comparison of elution time and absorbance spectra with commercial standards (Extrasynthese). The concentration of flavonols was calculated from the integrated absorbance of the two peaks at 353 nm.
Proanthocyanidins were extracted in 1/10 (w/v) skin/solvent suspension of 70% acetone in water (v/v) for 20 min in an Ultrasonic Cleaner bath. After centrifugation at 13,000 × g for 15 min, acetone was evaporated in a Speed Vac Concentrator (Savant) for 1 h. Water was added to obtain the initial volume. Proanthocyanidins were analyzed as reported in Harbertson et al. (2002). Absorbance was determined with a UV-160 spectrophotometer (Shimadzu). A standard curve was prepared using (+)-catechin. Proanthocyanidin values for skin extracts were obtained from the standard curve and reported in catechin equivalents.
Total RNA was extracted from 0.3 g of berry skin following the procedure described in Iandolino et al. (2004) and treated with 0.5 U/μg RQ1 DNase (Promega). First strand cDNA was synthesized using 2 μg of RNA, 0.5 μM (dT)18 primer, and 50 U of M-MLV Reverse Transcriptase (Promega). Quantitative real-time PCR was carried out in an ABI PRISM 7700 sequence detector (Applied Biosystems). Each reaction (20 μL) contained 250 nM of each primer, 5 μL of 1:400 diluted cDNA, and 10 μL of Power SYBR Green Master Mix (Applied Biosystems). Thermal cycling conditions were 95°C for 10 min followed by 95°C for 30 s, 58°C for 30 s, and 65°C for 60 s for 40 cycles, followed by a melting cycle from 65 to 95°C. Each cDNA sample was run in duplicate. Gene transcripts were quantified upon normalization to VvUbiquitin1 (TC32075, TIGR database) by comparing the cycle threshold (CT) of the target gene with that of VvUbiquitin1 (Bogs et al. 2005). Gene expression was expressed as mean and standard error calculated over all biological and technical replicates. Primers pairs for DFR, LDOX, and UFGT were retrieved from the literature (Goto-Yamamoto et al. 2002; FLS1 from Downey et al. (2003; LAR2 from Bogs et al. 2005; GST from Terrier et al. 2005). F3′H was amplified using the primer pair coded F3-1 in Castellarin et al. (2006) which amplifies all grape F3′H expressed in fruit among those deposited in NCBI (Bogs et al. 2006; Jeong et al. 2006). F3′5′H was amplified using the primer pair coded F35-1 in Castellarin et al. (2006) which amplifies the F3′5′H-1 isogenes of Castellarin et al. (2006) and the F3′5′H sequences of Bogs et al. (2006) and Jeong et al. (2006). F3H (Sparvoli et al. 1994) and BAN (Tanner et al. 2003) primers were newly designed by Castellarin et al. (2007).
Cumulative expression of genes throughout the ripening period was calculated as the area below the expression curves. Cumulative expression of genes of the core flavonoid pathway F3H, DFR and LDOX was calculated from 33 DAA to 133 DAA. The same was done for genes FLS1, BAN and LAR2 since flavonols and proanthocyanidin accumulated in the berries prior to veraison. Cumulative transcript levels of genes specific to anthocyanin biosynthesis, F3′H, F3′5′H, UFGT, and GST, were analyzed from 63 DAA to 133 DAA; the window of time corresponding to anthocyanin biosynthesis.
Differences between the treatments were tested for significance by one way randomized blocks analysis of variance using the COSTAT statistical package (CoHort software, CA, USA) and means were separated using Student–Newman–Keuls test.
Plant water status and berry growth
Berry growth in both C and LD vines before veraison was similar and significantly greater than in ED vines (Fig. 2b). Berries on C and LD vines reached maximum size at 96 DAA. At that time, C fruit were largest followed by LD fruit and then ED fruit. Water deficit applied before veraison strongly influenced berry weight throughout ripening. Nevertheless, ED vines resumed berry growth in the last 20 days before harvest and reached the same weight as the controls at harvest. LD caused the maximum reduction in berry size, particularly in the last period of ripening. The C and LD fruit lost about 6 to 8% of maximum fresh weight. Berries in ED vines did not reach maximum size until 125 DAA, when ED fruit were intermediate in size to C and LD fruit. At harvest berry weight was 1.11, 1.06, 0.969 g in C, ED, LD respectively. Yield per vine was 6.73, 6.48, and 5.62 kg of grapes vine−1, in C, ED, and LD respectively.
The transition from Stage II to Stage III based upon change in berry size was similar (77 DAA) for all treatments (Fig. 2b). The onset of sugar accumulation was also the same in all treatments (63 DAA) (Fig. 2c). The increase in Brix was rapid between 63 and 82 DAA and considerably slower after that. Water deficits influenced sugar concentration, and the influence depended upon the timing of the water deficit (Fig. 2c). Sugar accumulation was faster in ED vines than in C and LD vines between 63 and 82 DAA. Thereafter, increase in Brix was similar in C and ED vines but considerably faster in LD vines. At harvest, Brix values were 21.1, 23.4, and 25°Brix in C, ED, and LD vines, respectively.
However, the same transition based on color change was dependent on vine water status. Color change was accelerated in ED vines. At 67 DAA, 20% of berries on ED plants had already turned red, while no red berries were observed in C and LD treatments. At 74 DAA, 100% of ED berries were red, but only 70% were red in C and LD plants. By 81 DAA, all berries had turned red in all treatments.
ED vines started accumulating anthocyanins earlier than the other treatments; anthocyanins were easily detectable in ED at 67 DAA, while anthocyanin peaks just emerged from the background signal in C and LD berries (Figs. 3a, b). The maximum content of anthocyanins was attained 110 DAA in ED and LD vines, but not until 134 DAA in C vines.
Flavonoid pathway and berry development
The expression profiles of most structural genes corresponded with the observed changes in anthocyanin accumulation. The expression of genes common to the biosynthesis of anthocyanins, proanthocyanidins, and flavonols, F3H, F3′5′H, LDOX, and DFR, were induced by water deficit. In ED berries their expression was triggered earlier at veraison, and in general, remained greater than C berries for some time after the removal of water deficits (Fig. 4). LD treatment strongly enhanced gene expression in the last 20–30 days before harvest. In most cases expression returned to levels similar to C by 133 DAA.
Core flavonoid pathway genes and water deficits
The cumulative expression of anthocyanin pathway genes throughout development
Expression of F3′5′H encoding the hydroxylase responsible for the hydroxylation of the 5′ position on the B ring of the flavonoid skeleton drastically increased at 67 DAA (Fig. 4). Again, expression increased 1 week earlier in ED than C berries. After veraison, up-regulation of F3′5′H in ED persisted for approximately 1 month. Transcript levels were higher in LD in the middle of the ripening period, but the differences were statistically significant only at 95 DAA. Cumulative expression of F3′5′H was higher in both deficit treatments compared to C, and was greatest in LD berries (Table 1).
DFR was up-regulated in ED vines at 63 DAA, while in C berries the expression remained low until 74 DAA (Fig. 4). Differences were statistically significant at 67 DAA. Expression was greater in both ED and LD berries throughout the middle of the ripening period; 95, 109 and 133 DAA.
The expression profile of LDOX was similar between treatments until the completion of veraison, but from 95 DAA to harvest, LDOX expression was higher in LD than in C and ED with statistically significant differences at all sampling points except at 95 DAA (Fig. 4). Cumulative expression was 61.0, 71.5 and 84.2 in C, ED and LD, respectively (Table 1).
Transcripts of F3′H constantly increased in all treatments throughout ripening. The maximum level of the transcripts was reached at harvest for C and ED, whereas in LD berries the peak of expression pointed to 10 days before. Expression patterns in different treatments overlapped, except for 123 DAA when expression in LD berries was greater, but not statistically significant, than in C and ED berries. Cumulative expression of F3′H was approximately equal across all treatments (Table 1).
Response of anthocyanin biosynthesis and storage genes to water deficits
Genes strictly related to anthocyanin biosynthesis and storage, UFGT and GST, had similar patterns of expression (Fig.4). Transcripts were undetectable in green berries, and expression was triggered at the onset of veraison, and as the ripening process progressed, transcript levels greatly increased and remained at high levels. Expression was higher in ED and LD berries, from 95 to 123 DAA, but differences were statistically significant only for UFGT at 95 DAA. At 133 DAA, expression of both the genes decreased to C levels. Cumulative expression throughout ripening confirmed a strong up-regulation of UFGT and GST under water deficit (Table 1).
Regulation of flavonoid hydroxylases and the anthocyanin profile
The ratio between tri- and di-hydroxylated anthocyanins reflected the changes observed in the ratio of F3′5′H to F3′H expression, increasing from 67 to 123 DAA and finally decreasing at harvest (Fig. 6b). All treatments had the same trend but ED berries synthesized a higher percentage of tri-hydroxylated anthocyanins than C berries throughout ripening. These differences were statistically significant at all dates except 133 DAA. In LD berries this ratio was higher than in C berries only in the last phase of ripening, from 123 to 133 DAA. There were no differences between treatments in the concentration of di-hydroxlated anthocyanins; increases in the ratio resulted solely from increases in the tri-hydroxylated forms (data not shown).
Flavonol and proanthocyanidin accumulation
Total flavonols per berry increased in all treatments from the beginning of berry development till the end of ripening, but the concentration of flavonols, whether expressed per gram of skin or per g of berry, did not change except for a sharp increase during the last 10 days of ripening (data not shown). The flavonol concentration was 19% higher and 20% lower than controls at harvest in ED and LD fruit, respectively. These differences were not statistically significant. FLS1 transcripts were low for all treatments till the end of veraison (81 DAA), except for a peak of expression that occurred at 63 DAA (Fig. 4). From 95 DAA, FLS1 expression increased in both ED and LD treatments. In C berries expression did not increase significantly until 109 DAA and did not reach levels equal to those of the water deficit treatments until 133 DAA.
The results show that water deficits accelerated anthocyanin accumulation, particularly tri-hydroxylated anthocyanins, and increased the expression of many genes responsible for the biosynthesis of anthocyanins, including F3H, DFR, LDOX, F3′5′H, UFGT, GST. The consequences of early and late deficits were similar in the rapid accumulation of anthocyanins, transient expression pattern that favors tri- versus di-substituted anthocyanins, and limited effects on other flavonoids. The consequences were different in that ED had greater effects than LD, and ED advanced the onset of anthocyanin biosynthesis. The higher amount of anthocyanin per berry in ED fruit and the up-regulation of pertinent biosynthetic genes make clear that biosynthesis was increased and that higher concentrations of anthocyanin in ED fruit was not simply the result of reduced berry growth.
Early water deficits accelerated the shift to anthocyanin biosynthesis and increased the initial rate of sugar accumulation, but water deficits did not accelerate all the changes associated with the onset of ripening. Color change is one of the most universal and easily recognizable aspects of the ripening process. The changes in the flavonoid pathway that bring about this are part of a larger coordinated change in berry development at the onset of ripening that includes growth, softening, and sugar accumulation. Genes strictly related to anthocyanin biosynthesis and storage, UFGT and GST, were activated earlier in ED berries than in controls along with the core flavonoid pathway genes needed for anthocyanin biosynthesis. The accelerated increase in anthocyanins and anthocyanin-related gene expression caused by ED suggests that water deficits accelerated the onset of ripening, i.e. veraison. The earlier appearance of red berries in the field is consistent with this suggestion.
However, several lines of evidence argue that ED affected anthocyanin synthesis per se and not the suite of changes that comprise ripening. In contrast to the accelerated accumulation of anthocyanins, there was no effect of ED on the onset of Stage III growth, which occurred almost 2 weeks later. Also, the onset of rapid sugar accumulation was coincident in all treatments, although the Brix was higher in ED berries early in development. These same responses to early water deficits, i.e., no change in timing of growth or sugar accumulation but an acceleration of anthocyanin accumulation, were observed in Cabernet Sauvignon (Acevedo et al. 2004) and in Merlot (Castellarin et al. 2007), but no effect on anthocyanins was observed in Shiraz (Ojeda et al. 2002). Also, if the ripening process was accelerated in ED, an earlier down-regulation of the genes responsible for proanthocyanidins would be expected. There was no evidence for acceleration of the loss of BAN and LAR2 expression or of proanthocyanidin accumulation. Thus, in several studies the onset of growth and sugar accumulation aspects of ripening were not advanced by water deficits. Even within grapevine, water deficits reportedly accelerate (Keller 2005) and delay (Sipiora and Granda 1998) fruit maturation.
ABA increases at veraison and is widely used to promote ripening in grapes. For example, exogenous ABA increased expression of VvmybA1, expression of related genes for anthocyanin synthesizing enzymes, and the concentration of anthocyanins (Jeong et al. 2004). ABA also activates invertases (Pan et al. 2005) and in grape berries monosaccharides accumulate, not sucrose. A monosaccharide transporter homologue (VvHT1, Vitis vinifera hexose transporter 1) has been isolated from grape berries at veraison, and its expression was shown to be regulated by sugars and ABA (Cakir et al. 2003). Presumably the hexoses involved are imported as sucrose via the dramatic increase in phloem influx at veraison (Greenspan et al. 1994, 1996). The transition to phloem influx and anthocyanin accumulation is accompanied by a decrease in fruit cell turgor (Thomas et al. 2006) that may be required for sink accumulation of phloem sap (Patrick 1997). Thus, preveraison water deficits, which lower turgor even more (Thomas et al. 2006), may cause increases in sugar influx and ABA, which in turn activate the expression of sugar transporters and anthocyanin pathway genes. However, grape may be unusual in that when ABA-related genes commonly induced by water deficits (NCED1, NCED2, and rd22) were assayed in berry skin, water deficits only transiently increased their expression at veraison (Castellarin et al. 2007).
Water deficits had only a mild effect on proanthocyanidin biosynthesis and accumulation. ED enhanced the expression of LAR2 before veraison, but these changes were minor. During ripening there were no statistically significant differences in mean proanthocyanidin concentrations, although they were greater in both early and late water deficits and similar in magnitude to differences reported by Ojeda et al. (2002). Kennedy et al. (2002) also showed that proanthocyanidin production is enhanced by water deficit. In this study, the difference in proanthocyanidin concentration essentially disappears when the data are expressed on a per berry basis (data not shown).
These results have implications for winegrape production. Water deficits had direct effects on flavonoid gene expression and metabolism. Therefore, differences in fruit composition were not attributable simply to effects of water deficits on berry growth (Roby et al. 2004). Sugars and anthocyanins accumulated more rapidly in ED and LD in comparison to C vines. Maximum anthocyanin content was attained 24 days earlier in ED and LD. Maintaining these fruit on the vine beyond this peak of anthocyanin content resulted in some loss of color. Thus, although water deficits did not advance the onset of ripening, the data suggest that ED and LD accelerated fruit development to maturity, and both LD and ED fruit could be harvested before C fruit and have the same or higher concentrations of sugars, anthocyanins, and proanthocyanidins. The increase in 3′4′5′-hydroxylated anthocyanins constitutes an enrichment of purple/blue pigments, hence modifying grape and must color (Castellarin et al. 2006). This modification puts more of the pigments in moieties that resist oxidation and resultant color change (Sarni et al. 1995).
Finally, fewer significant differences were observed in the expression data than in other assays. There were significant treatment effects on plant water status, berry size, and concentrations of sugars and anthocyanins on essentially every sample date. In contrast, gene expression assays revealed far fewer statistically significant differences. Although mitigated by the observed differences in means maintained over the time course, this contrast cannot be resolved without further work, but has important implications for interpretation in similar studies. We can argue that analysis of gene expression results inherently in more noise than is produced in plant water status, sugar concentration, etc. by the vagaries of the field environment. This may be compounded by the reality that ripening is asynchronous among berries within clusters, especially at veraison, making samples rather heterogeneous. Therefore the variance between biological replicates may reflect this biological heterogeneity. More rather than less replication may be important for accurate insight into the responses of gene expression to environmental conditions in the field.
The authors thank R.H. Phillips for cooperation and access to vineyard, Mark Krasnow for field assistance, Greg Giguere for water management, and Mark Downey for advice in HPLC analysis.